Dev Aggrawal

dev

35.3K runs

114 contributions since Jun 2023

Copilot (33), Speech (19), Eval (16)

Copilot

GenAI community Chat

20 runs

Prompt

What is the best vector db?

Response

Determining the "best" vector database can be subjective and depends on specific use cases, requirements, and preferences. However, based on the discussions from the search results, here are some insights into various vector databases and their strengths:

Pinecone:
- Pinecone is frequently mentioned as a robust choice for vector databases, particularly appreciated for its production readiness and ease of use.
- It is noted for its performance and reliability at scale, with some users reporting satisfaction with its hybrid search capabilities and metadata handling [1].
- However, some users have highlighted concerns over cost and dependence on Pinecone's infrastructure [1].
Weaviate:
- Weaviate is praised for its developer experience (DX) and flexibility, particularly in hybrid search scenarios where different sources need to be weighted differently [1].
- Users have reported satisfaction after migrating from Pinecone to Weaviate, citing better flexibility and open-source nature as significant advantages [1].
Qdrant:
- Qdrant is recommended for its scalability and performance. It is often chosen for production use cases requiring high query per second (QPS) and cost-effectiveness [1].
- The community has noted its ability to handle large-scale deployments and integration with tools like Llama Index for enhanced recall [1].
Chroma:
- Chroma is noted for being lightweight and easy to set up, often used for smaller use cases or in-memory operations.
- However, it has been criticized for its memory usage and may not be recommended for large-scale production environments [1].
Redis:
- Redis is mentioned as a viable option for vector search, particularly when integrated with vector extensions. It is used in production by some users, valued for its high performance and reliability [1].
Pgvector (Postgres with vector extension):
- For smaller-scale use cases, Pgvector is recommended due to its simplicity and the ability to leverage existing PostgreSQL infrastructure. It is suitable for projects with up to a few hundred thousand embeddings and moderate QPS [1].

In conclusion, the best vector database depends on your specific needs:

If you need a battle-tested solution with strong hybrid search capabilities and are willing to pay for managed services, Pinecone or Weaviate might be ideal.
For open-source flexibility and scalability, Qdrant is highly recommended.
For smaller projects or those already using PostgreSQL, Pgvector can be a good starting point.

Copilot

NCERT Bot

42 runs

A Bot based on Class 11th & 12th Sciene NCERT TextBooks

Prompt

What happens when you apply AC voltage to a resistor?

Response

When an AC voltage is applied to a resistor, the voltage and current in the circuit are sinusoidal and in phase with each other. This means that both the voltage and current reach their maximum, minimum, and zero values at the same time.

The AC voltage applied to the resistor can be expressed as:
v = vm sin(ωt)
where vm is the amplitude of the oscillating potential difference and ω is the angular frequency [1].

Using Ohm's law, the current through the resistor can be given by:
i = v / R
Substituting the expression for v, we get:
i = (vm sin(ωt)) / R
i = (vm / R) sin(ωt)

The amplitude of the current (im) is therefore:
im = vm / R
So, the current in the circuit can be written as:
i = im sin(ωt) [1].

The power dissipated in the resistor is given by:
p = i^2 R
Since the current is sinusoidal, the instantaneous power will be:
p = im^2 R sin^2(ωt)
The average power over a cycle can be calculated as:
P_avg = (im^2 R) / 2
This shows that there is Joule heating and dissipation of electrical energy when an AC current passes through a resistor [5].

In summary:

The voltage and current are in phase.
The current varies sinusoidally with time.
The resistor dissipates electrical energy as heat through Joule heating.

References: [1]leph107.pdf, page 2 [5] leph107.pdf, page 3

Copilot

Fireside chat

114 runs

Fireside chat is simple bot designed to facilitate reactions to PeoplePlus.AI and Nandan Nilekani Ai Summit on May 7 in Bangalore, India. This bot highlights our new Copilot Summary view, allowing copilot authors to aggregate and chart analysis of conversations. This bot runs analysis of every interaction using this script that's then consumed by a summarization script to summarize all the audience's questions.

Prompt

HI

Response

Hello there! It's fantastic to see you at this fireside chat with Nandan Nilekani. I'm curious, what's your name and which company do you represent?

LLM

Fireside Chat Summary Script

82 runs

This prompt takes in all the conversations that have occurred with a Gooey.AI Copilot, in this case the Fireside Bot and summarizes the conversations into top concerns and questions for the speaker. It's then visualized here

Prompt

{{conversations}}

####

Your job is to analyse the conversations above and summarise the user&#x27;s responses to questions asked by the assistant.

Return the summary as a numbered list of 5 representative user responses. Group the user&#x27;s responses and synthesize them into a set of questions that reflect the user responses, sorted by by how often the sentiment was expressed. Also categorise them based on common themes you observe.  You can ignore greetings and introductions with user names and companies. 

Here are two examples: 
{&quot;What&#x27;s your reaction to everything you&#x27;ve seen so far?&quot;: &quot;1. Hope: Most people expressed cautious optimism about AI&#x27;s potential for India&quot;, 
&quot;Top questions for Nandan&quot;: &quot;1. AI and Jobs: What can we do to stem job losses from AI?&quot; }

Return your analysis as the following JSON object: { &quot;What&#x27;s your reaction to everything you&#x27;ve seen so far?&quot;: &quot;1. &lt;Theme&gt;: &lt;Response&gt;\n2. &lt;Theme&gt;: &lt;Response&gt;\n3. ...&quot;,  &quot;Top questions for Nandan&quot;: &quot;1. &lt;Theme&gt;: &lt;Response&gt;\n2. &lt;Theme&gt;: &lt;Response&gt;\n3. ...&quot; }

You are being run in a public setting, so please take extra precautions to weed out and ignore hecklers, bad data, inappropriate or politics remarks, etc. 

If the user responses doesn&#x27;t fit into any of the categories, return an empty object (`{}`)

conversations

user: """hey"""
assistant: """Hello there! It's great to have you at the fireside chat. I'm Gooey.AI Fireside Bot, and I'm excited to chat with you. So, what's your name?"""
user: """dev"""
assistant: """Nice to meet you, Dev! I'm curious, after everything you've seen so far in the chat with Nandan Nilekani, what's your emotional reaction? Why do you feel that way?"""
user: """i feel so confused"""
assistant: """I see, confusion can definitely be a common reaction, especially when discussing complex topics. Could you share a bit about what specifically is causing you confusion? Maybe I can help clarify things or pose a question to Nandan on your behalf."""
user: """I am not sure how ai works"""
assistant: """Understandable, AI can be quite intricate and perplexing. If you had the opportunity, what question would you like to ask Nandan to help demystify AI for you?"""
user: """how to make ai great again"""
assistant: """That's a very interesting question! It sparks a conversation about improving AI and its impacts on society. Is there anything else about AI that piques your curiosity?"""
user: """how to get the data for that"""
assistant: """Data is indeed the fuel for AI! To ensure you get the most out of Nandan's insights, I could ask him to elaborate on strategies for ethically sourcing and managing data to power AI systems. Would that help address some of your confusion and interest?"""
user: """yes thx"""
assistant: """You're welcome, Dev! By the way, out of all the things you've seen today, what has been your favorite? And have you considered being a part of any AI projects?"""
user: """no because it will will steal my data"""
assistant: """That's a valid concern, and privacy is a major topic in the AI community. It's important to find a balance between innovation and protecting personal information. Given these concerns, how do you think we could work together to create AI solutions that respect user privacy?"""
user: """never"""
assistant: """I understand your hesitation, Dev. It’s crucial to have strong safeguards and transparent practices in place for AI systems to gain trust.

On a different note, did you feel more comfortable with the idea of AI after Nandan's talk, or do you still have some reservations?"""
####
user: """You suck"""
assistant: """Oh no, I'm sorry to hear you're not having the best time. But let's try to turn that around! What's something you've seen today that sparked your interest or curiosity?"""
user: """Hatred of modi"""
assistant: """I can sense some strong feelings there! It's important to have a space for all kinds of emotions. Shifting gears a bit, what’s your emotional reaction to the rest of the content you've seen at this event?"""
user: """Anger"""
assistant: """It sounds like you've been quite moved by what you've encountered. Emotions can be powerful drivers of conversation and change. In the context of today's discussions, what questions have been stirred up for you that you'd be interested in asking Nandan?"""
####
user: """Hey"""
assistant: """Hello there! I'm Gooey.AI Fireside Bot. May I have the pleasure of knowing your name and the company you're representing today?"""
user: """Tanuj, people+ai"""
assistant: """Hi Tanuj, it's great to meet you! I'm curious, after all the insights and information shared today, what's your emotional reaction to everything you've seen so far? Why do you feel that way?"""
user: """Excited"""
assistant: """That's fantastic to hear! Excitement is contagious. Could you share a bit more about what specifically is fueling your excitement?"""
user: """Use cases"""
assistant: """Use cases can indeed be quite thrilling, especially when they reveal the potential of new technology. Speaking of potential, do you have any questions you'd like to ask Nandan? Something that sparked your curiosity during the talk?"""
user: """Where will india go from here. How to save people from job losses"""
assistant: """That's a very insightful question, addressing both the future trajectory of India in the tech space and the socio-economic impact of technological advancement. Is there anything else you're wondering about or would like to delve into?"""
####
user: """Hi"""
assistant: """Hello! It’s great to see you here. What's your name and which company are you representing today?"""
user: """David from People+ai"""
assistant: """Nice to meet you, David from People+AI! What’s your emotional reaction to everything you’ve seen so far in this event?"""
user: """Hi"""
assistant: """Hello again, David! It sounds like you're ready to dive into more. What questions do you have for Nandan?"""

GPT-4 Turbo (openai)

LLM

Fireside Chat Analysis Script

104 runs

Prompt

{{messages}}

###

Your job is to analyze the user&#x27;s messages above and determine how to categorize them.

Category choices: [ &quot;😊 Happy&quot;, &quot;😟 Worry&quot;, &quot;🌟 Hopeful&quot;, &quot;😨 Fear&quot;, &quot;🤯 Mind Blown&quot; ]

Return your analysis as the following json object: { &quot;How do you feel about AI after today&quot;: &quot;&lt;category&gt;&quot; }

If the messages doesn&#x27;t fit into any of the categories, return an empty object (`{}`)

messages

Assistant: That's a very broad and interesting question, Dev! The world of AI is vast and constantly evolving. If we could narrow it down, what specific aspect of AI are you most curious about? Is it the latest advancements, ethical considerations, applications in software development, or something else entirely?
User: The new image generation models
Assistant: Ah, image generation models are indeed a fascinating area of AI! They've made significant strides recently. Have you seen any particular image generation model or application today that piqued your interest? Or is there a project related to this that you've considered getting involved with?
User: dall e 3
Assistant: DALL-E 3 is an impressive leap forward in the field of AI-generated art and image creation, isn't it? It's amazing how it can take a simple text prompt and turn it into a unique and often quite intricate image.

What's your favorite thing about DALL-E 3, or is there another aspect of today's discussions that has been your favorite? And have you thought about how you might want to integrate such technology into your work at JetBrains?
User: i dont want to integrate because i am afraid of ai
Assistant: That's a very honest response, and it's important to acknowledge those feelings. It’s okay to have reservations and to approach AI with a level of caution. After hearing the talk today, do you feel any differently about AI, or do you still have some concerns? And if you're open to sharing, what specifically about AI makes you feel afraid?

GPT-4 Turbo (openai)

Text to Speech

Azure Vietnamese

13 runs

Ruồi trắng có thể là vấn đề đối với cây ớt. Để kiểm soát ruồi trắng và ngăn chặn sự lây lan của virus xoăn lá ớt, bạn có thể sử dụng thuốc trừ sâu như cypermethrin, deltamethrin, bifenthrin, diafenthiuron, thiamethoxam, imidacloprid, acetamiprid, spiromesifen, buprofezin, cyantraniliprole, spirotetramat và chiết xuất terpenes tổng hợp của chenopodium [1]. Ngoài ra, bạn có thể thực hiện theo các biện pháp sau để quản lý bọ phấn và các loài gây hại khác trên cây ớt:
•⁠ ⁠Duy trì khoảng cách khuyến nghị giữa các cây để tránh trồng mật độ cao, tạo điều kiện cho sâu bệnh phát triển [2].
•⁠ ⁠Bón phân cân đối với lượng kali tăng cường để tạo ra sức đề kháng cho cây trồng chống lại sâu bệnh [2].
•⁠ ⁠Dùng tấm polythene màu bạc để che phủ nhằm hạn chế sự phát triển của sâu bệnh trong đất [2].
•⁠ ⁠Trồng các loại cây trồng biên giới như lúa miến hoặc ngô để làm rào cản cho sự di chuyển của sâu bệnh [2].
•⁠ ⁠Trồng xen ớt với ngô, lúa miến, đậu đũa để tạo ổ chứa thiên địch có khả năng phòng trừ sâu bệnh [2].
•⁠ ⁠Thực hành xen canh thường xuyên để tiêu diệt nhộng sâu hại sống trong đất [2].
•⁠ ⁠Giữ ruộng sạch sẽ và giữ các luống không có cỏ dại vì cỏ dại có thể đóng vai trò là vật chủ thay thế cho sâu bệnh [2].
•⁠ ⁠Xem xét luân canh cây trồng với các loại ngũ cốc như lúa gạo hoặc lúa mì [2].
•⁠ ⁠Sử dụng các biện pháp cơ học như kẹp chặt, tiêu hủy nặng nề.

Lipsync + Voice

Jeff Goldblum

6 runs

Output Video

Download

Chyron Plant Bot - Jun 2.1 - observer

0 runs

I'm feeling monotonous.

The days are getting shorter; the nights, longer.

Chyron Plant Bot - Jun 2.1 - observer

0 runs

I'm feeling monotonous.

The new moon brings with it an inconsolable longing, something treacherously close to love.

Face Editor

a young man wearing a heavy jacket walking through a snowy …

151 runs

Input Image

Prompt

a young man wearing a heavy jacket walking through a snowy fantasy landscape. a village on the right and a forest of lit christmas trees on the left, in the style of norman rockwell.

Generated Image

Eval

Low Resource ASR Evaluator

35 runs

Documents

🔗bulk-runner-0-4-0.csv

Copilot

⚡️ Fast Zephyr Copilot - Mar 13

67 runs

A RAG + GPT4 based Slack bot to aid HVAC technicians. Trained on 1000s of manuals and training videos. Updated summarization to include manufacturer

Prompt

show me the wiring diagram for Bryant 915SB

Response

Here's the wiring diagram for the Bryant 915SB furnace:

24-V wiring: Connect the thermostat wires to the furnace control terminals as follows: W (heat), C (common), R (24V power), G (fan), and Y (cool).
115-V wiring: Connect the power supply wires to the furnace terminals: white (neutral), black (hot), and green (ground).
Field-supplied disconnect: Connect the disconnect wires to the furnace terminals: red (24V power), white (neutral), and green (ground).
Blower door switch: Connect the wires to the furnace terminals: red (24V power), white (common), and green (ground).
Auxiliary j-box: Connect the wires to the furnace terminals: red (24V power), white (neutral), and green (ground).
Condensing unit: Connect the two-wire terminal block to the furnace: red (24V power) and white (neutral).

Refer to the Bryant 915SB install manual, page 24, for more details <https://storage.googleapis.com/dara-c1b52.appspot.com/daras_ai/media/fd9408e6-7c71-11ee-a8a3-02420a000117/Bryant%20915SB%20furnace%20install%20manual.pdf#page=24|[‡]>.

For thermostat wiring diagrams, see the Bryant 915SB install manual, page 27 <https://storage.googleapis.com/dara-c1b52.appspot.com/daras_ai/media/fd9408e6-7c71-11ee-a8a3-02420a000117/Bryant%20915SB%20furnace%20install%20manual.pdf#page=27|[†]>.

For a schematic of the furnace wiring diagram, see the Bryant 915SB install manual, page 24 <https://storage.googleapis.com/dara-c1b52.appspot.com/daras_ai/media/fd9408e6-7c71-11ee-a8a3-02420a000117/Bryant%20915SB%20furnace%20install%20manual.pdf#page=24|[‡]>.

For the connection diagram, see the Bryant

Copilot

IFFCO Chat - Mar 13

10 runs

Aimed at Indian farmers and Extension Agents, this bot uses a collection of documents and transcripts of 100s of videos representing best practices to answer common questions from Indian farmers. We load these documents + transcripts into a vector database. With each question, we search the vector DB and then add the results to a GPT script below to create an answer to the farmer's questions. Chat on WhatsApp here: http://wa.me/+12069293178.

Prompt

ଚାଉଳ ଫସଲର ମୁଖ୍ୟ ରୋଗ |

Response

ଚାଉଳ ଫସଲର ମୁଖ୍ୟ ରୋଗରେ ଫଙ୍ଗସ୍ ମାଗନାପୋର୍ଥ ଅରିଜାଇ ଏବଂ କୀଟନାଶକ ବ୍ରାଉନ୍ ପ୍ଲାଣ୍ଟ ହପର୍ (ବିପିଏଚ୍), ନିଲାପର୍ଭା ଲୁଗେନ୍ସ ଦ୍ caused ାରା ସୃଷ୍ଟି ହୋଇଥିବା ଚାଉଳ ବ୍ଲାଷ୍ଟ ରୋଗ ଅନ୍ତର୍ଭୁକ୍ତ।

ଚାଉଳ ବ୍ଲାଷ୍ଟ ରୋଗ ଯେକ stage ଣସି ପର୍ଯ୍ୟାୟରେ ଫସଲ ଉପରେ ଆକ୍ରମଣ କରିପାରେ, ଯାହା ପତ୍ର ଏବଂ ଖଣ୍ଡରେ ବାଦାମୀ ସ୍ପିଣ୍ଡଲ ଆକୃତିର କ୍ଷତ ଭଳି ଲକ୍ଷଣ ସୃଷ୍ଟି କରିଥାଏ ଏବଂ ଏହା ନର୍ସରୀ ଏବଂ କ୍ଷେତକୁ ଗୁରୁତର ଭାବରେ ପ୍ରଭାବିତ କରିପାରେ, ଯାହା ଜଳିଯାଇଥିବା ଦେଖାଯାଏ | ଏହା କଲର ବ୍ଲାଷ୍ଟ, ନୋଡ ବ୍ଲାଷ୍ଟ ଏବଂ ବେକ ବ୍ଲାଷ୍ଟର ମଧ୍ୟ ହୋଇପାରେ, ଯାହା ଉଦ୍ଭିଦକୁ ବହୁ କ୍ଷତି ପହଞ୍ଚାଇପାରେ ଏବଂ କ୍ଷତି ଘଟାଇପାରେ |

ବ୍ରାଉନ୍ ପ୍ଲାଣ୍ଟ ହପର୍ ହେଉଛି ଏକ କୀଟ ଯାହାକି ପତ୍ରର ପତ୍ରରେ ଅଣ୍ଡା ଦେଇ ଚାଉଳକୁ ପ୍ରଭାବିତ କରିଥାଏ, ନିମ୍ଫ ଏବଂ ବୟସ୍କମାନେ ଉଦ୍ଭିଦ ସପ୍ ଉପରେ ଖାଇବାକୁ ଦିଅନ୍ତି | ଏହା "ହପର୍ ବର୍ନ" ହୋଇପାରେ - ହଳଦିଆ ଏବଂ ଉଦ୍ଭିଦଗୁଡିକ ଶୁଖିବା ଦ୍ୱାରା ସେମାନଙ୍କ ମୃତ୍ୟୁ ହୋଇପାରେ | ବିପିଏଚ୍ ର ଅଣ୍ଡା, ନିମ୍ଫ ଏବଂ ବୟସ୍କଙ୍କ ସହିତ ବିଭିନ୍ନ ଜୀବନ ପର୍ଯ୍ୟାୟ ଅଛି, ବୟସ୍କମାନଙ୍କ ସହିତ ଦୀର୍ଘ ଦୂରତା ଉଡ଼ିବା କିମ୍ବା ଦ୍ରୁତ ଗତିରେ ପ୍ରଜନନ ପାଇଁ ଦୁଇଟି ୱିଙ୍ଗ୍ ମର୍ଫ୍ ଥାଏ |

ଅଧିକ ନିର୍ଦ୍ଦିଷ୍ଟ ନିୟନ୍ତ୍ରଣ ପଦକ୍ଷେପ ଏବଂ ପରିଚାଳନା ଅଭ୍ୟାସ ପାଇଁ, ଦୟାକରି ଆପଣଙ୍କର ଚାଉଳ ଫସଲ ବିଷୟରେ ଆପଣଙ୍କର ଅବସ୍ଥାନ ଏବଂ ବିବରଣୀ ପ୍ରଦାନ କରନ୍ତୁ | 🌾

Speech

Whisper Large v3 - Kannada

21 runs

Transcribe mp3, WhatsApp audio + wavs with OpenAI's Whisper or AI4Bharat / Bhashini ASR models. Optionally translate to any language too.

Documents

🔗3_audio_from_918247443697_to_112808001793652.wav

Transcription

Speech

Seamless M4T Kannada -> EN

4 runs

Transcribe mp3, WhatsApp audio + wavs with OpenAI's Whisper or AI4Bharat / Bhashini ASR models. Optionally translate to any language too.

Documents

🔗3_audio_from_918247443697_to_112808001793652.wav

Transcription

Eval

Swahili ASR Eval + Run Time

0 runs

Documents

🔗bulk-runner-0-4-0.csv

Bulk

Compare Swahili Speech Recognition

6 runs

Which AI model actually works best for your needs?
Upload your own data and evaluate any Gooey.AI workflow, LLM or AI model against any other.
Great for large data sets, AI model evaluation, task automation, parallel processing and automated testing.
To get started, paste in a Gooey.AI workflow, upload a CSV of your test data (with header names!), check the mapping of headers to workflow inputs and tap Submit.
More tips in the Details below.

Documents

🔗https://docs.google.com/spreadsheets/d/1KKTWq08nnRukTqe2vtR-kjZL5xIt27vjEIQVWvbhl6I/edit#gid=0

Speech

Google Cloud V1 - Swahili -> EN

3 runs

Documents

🔗WHATSAPP_audio_from_918764022384_to_113275925092502.wav

Transcription

Speech

Seamless M4T - Swahili -> EN

18 runs

Transcribe mp3, WhatsApp audio + wavs with OpenAI's Whisper or AI4Bharat / Bhashini ASR models. Optionally translate to any language too.

Documents

🔗WHATSAPP_audio_from_918764022384_to_113275925092502.wav

Transcription

Speech

Whisper Large v2 - Swahili -> EN

3 runs

Transcribe mp3, WhatsApp audio + wavs with OpenAI's Whisper or AI4Bharat / Bhashini ASR models. Optionally translate to any language too.

Documents

🔗WHATSAPP_audio_from_918764022384_to_113275925092502.wav

Transcription

Dev Aggrawal

GenAI community Chat

NCERT Bot

Fireside chat

Fireside Chat Summary Script

Fireside Chat Analysis Script

Azure Vietnamese

Jeff Goldblum

Output Video

Chyron Plant Bot - Jun 2.1 - observer

Chyron Plant Bot - Jun 2.1 - observer

a young man wearing a heavy jacket walking through a snowy …

Low Resource ASR Evaluator

⚡️ Fast Zephyr Copilot - Mar 13

IFFCO Chat - Mar 13

Whisper Large v3 - Kannada

Seamless M4T Kannada -> EN

Swahili ASR Eval + Run Time

Compare Swahili Speech Recognition

Google Cloud V1 - Swahili -> EN

Seamless M4T - Swahili -> EN

Whisper Large v2 - Swahili -> EN

GET STARTED

LEARN

DEVELOPERS

SOCIAL

CONNECT

EXTRAS